|
https://issues.apache.org/bugzilla/show_bug.cgi?id=53169
Priority: P2 Bug ID: 53169 Assignee: [hidden email] Summary: [patch] don't do chunking with Connection: close Severity: minor Classification: Unclassified OS: All Reporter: [hidden email] Hardware: All Status: NEW Version: trunk Component: Catalina Product: Tomcat 7 Created attachment 28702 --> https://issues.apache.org/bugzilla/attachment.cgi?id=28702&action=edit patch against trunk The attached patch disables chunking on if there is a Connection: close header, we're on HTTP 1.1 and there's no Content-Length header. This helps to implement Server-Sent Events [1]. Server-Sent Event are conceptually similar to the forever streaming iframe in the sense that there's an "infinite" response from the server that always gets updated. But they're easier to use and with less issues. They are also easier to use than WebSockets if you don't need a back channel. The specification reccoments to disable chunking. Jetty implements the same behavior, when there's a Connection: close header it doesn't do chunking. The following discussion [2] leads me to believe that patches would be welcome. The patch comes with unit tests that verify that there is still chunking happending when there is no Connection: close header. [1] http://dev.w3.org/html5/eventsource/#notes [2] http://tomcat.10.n6.nabble.com/How-to-disable-chunked-encoding-for-the-Http11NioProtocol-connector-td2038448.html -- You are receiving this mail because: You are the assignee for the bug. |
|
https://issues.apache.org/bugzilla/show_bug.cgi?id=53169
Mark Thomas <[hidden email]> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution|--- |FIXED --- Comment #1 from Mark Thomas <[hidden email]> --- The general approach is fine although I have made a number of changes to the patch: - The Servlets in the unit tests should exist in the test class that uses them - There is no need use the test webapp for these unit tests - Use more of the built-in infrastructure for making a request - Improved the checking of the response headers These changes identified that the Connection: close header was being set twice. It isn't illegal but it might confuse some clients. I added some additional code to prevent this. The change has been made to trunk and 7.0.x and will be included in 7.0.28 onwards. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [hidden email] For additional commands, e-mail: [hidden email] |
|
In reply to this post by Bugzilla from bugzilla@apache.org
https://issues.apache.org/bugzilla/show_bug.cgi?id=53169
--- Comment #2 from Philippe Marschall <[hidden email]> --- Thanks for cleaning up an applying. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [hidden email] For additional commands, e-mail: [hidden email] |
| Powered by Nabble | Edit this page |
